Want to protect your cyber security and still get fast solutions? Ask a secure question today.Go Premium

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 677
  • Last Modified:

Trunk link between HP 9300 (Foundry) and Dell Poweredge server

Hi Guys,

I am having a few problems setting up trunked links to a couple of servers. I have an HP 9308 running 07.8.00a (This is v. similar to a foundry device) and we are using Dell Poweredge servers running Windows 2003 enterprise.

The servers are setup with 2 internal network cards and these connections are 'bridged' from within windows. (NOT using any 3rd party Software)

How should i configure the switch in order to make this function correctly? Do i need to set up a 'Server' trunk? Or Not? I am unsure how Windows itself configures the trunk....

Any help much appreciated,

Cheers
0
ian_chard
Asked:
ian_chard
  • 2
1 Solution
 
pseudocyberCommented:
When you say "trunk" we assume you mean port aggregation, right?  Combine the two interfaces for "faster" throughput?

Do NOT bridge the two NICs together.  If you do that with the Windows OS and then plug them into your switch, you just created a loop which can take down your network.  If you're running Spanning Tree on the HP's they probably detected that and shut down one of your ports.

You don't want to use anything on the OS to do this.  

You do want to use the NIC drivers to do it.  Not sure about Dells, but you probably (or you need) some kind of Teaming software.  

There are three types:
1.) Transmit load balancing - good for a server sending to many different devices, as long as it has many interfaces.  Typically a web server, or a file server which doesn't receive much.  All receive comes in one one port.  All transmit is statistically split over the other ports.

2.) Fault Tolerant Load Balancing - two or more connections are plugged in and ready.  Only one is active.  One becomes unplugged or goes down, traffic moves over to one of the other ones.

3.) Switch Assisted Load Balancing - two or more connections are plugged in and ready.  All interfaces are active and all interfaces are statistically load balanced so that they all transmit and receive.

NOW NOTE THIS:  Traffic from destination A to destination B will ALWAYS use the SAME INTERFACES - SAME PATH as long as everything is up.  So, if your aim is to achieve greater throughput between two servers - you need a GigE interface.  This is the way it works - some algorithm will look at the source and destination and make a mathematical decision on which interface to use - it will always use the same interface, because of the math.  THE FLOW IS NOT BROKEN UP BETWEEN INTERFACES.

So, if you're many to many - you're good.  If you're many to one, you're good.  If you're one to one - using load balancing to achieve greater throughput won't work.  However, if your aim is to achieve better resilience - fault tolerance then it's a good idea.

To achieve Switch Assisted Load Balancing - you have to set it up on the server and the switch.  On the switch side, it's called link aggregation.  You can use LACP (Link Aggregation Control Protocol) which is open standard.  Or you can use a proprietary method according to the vendor of the network gear - Cisco uses Fast Ether Channel, Nortel uses Multi Link Trunking.

So, if you still need help, let us know what kind of Teaming you want to do, if you have the Teaming drivers on your server, and what your switch can do as far as Link Aggregation goes.

Hope this helps.
0
 
ian_chardAuthor Commented:
Thanks for that, that was a really good clear explanation. Apologies for not closing the question earlier.

Thanks again
Ian
0
 
pseudocyberCommented:
You're welcome.  Thanks.  :)
0

Featured Post

Concerto's Cloud Advisory Services

Want to avoid the missteps to gaining all the benefits of the cloud? Learn more about the different assessment options from our Cloud Advisory team.

  • 2
Tackle projects and never again get stuck behind a technical roadblock.
Join Now